This K08 proposal will develop the career of Dr. James Floyd, a general internist with training and interests in cardiovascular disease epidemiology and pharmacoepidemiology. His goals are: 1) to evaluate the cardiovascular effects of various combinations of diabetes therapies; and 2) to become a successful independent investigator studying cardiovascular drug safety. These goals will be accomplished through an intensive mentored research experience in the design and conduct of cardiovascular drug safety studies and training in the use of several epidemiologic methods that can be used to address confounding and other forms of bias common in observational studies. These activities will take place under the supervision of primary mentor Dr. Bruce Psaty and co-mentors Dr. Barbara McKnight and Dr. Miguel Hernan, all of whom are accomplished investigators with expertise in cardiovascular disease, epidemiologic methods, and the design and conduct of drug safety studies. Insulin and oral glucose-lowering drugs are leading causes of hospitalizations for adverse drug effects in the U.S., and about 40% of persons with type 2 diabetes mellitus (DM) on insulin therapy use oral agents in combination. Concerns about the cardiovascular safety of oral diabetes therapies first emerged over forty years ago, when the first-generation sulfonylurea tolbutamide was found to increase the risk of cardiovascular death in the University Group Diabetes Program clinical trial. Sulfonylureas may cause adverse cardiovascular events as a consequence of hypoglycemia and through direct myocardial effects. We hypothesize that among persons with type 2 DM using long-acting insulin, the use of sulfonylureas compared with the use of long- acting insulin alone increases the risk of cardiovascular disease and death. In the setting of Group Health Cooperative, a large integrated healthcare delivery system, we have previously used electronic prescription records and health data obtained from medical records to conduct population-based studies of the cardiovascular safety of other classes of medications. Leveraging existing data from these studies, we propose to evaluate the cardiovascular safety of sulfonylureas and other glucose-lowering therapies in two linked epidemiologic studies: (1) an MI case-control study of long-acting insulin users with type 2 DM and no prior history of MI; and (2) an inception cohort study of survivors from the MI case group of the first study. Restriction to subjects without prevalent cardiovascular disease, adjustment for key potential confounding factors, analytic methods such as propensity score calibration, and sensitivity analyses will be used to address residual confounding, the most serious threat to the validity of these studies. Causal inference methods will be used to address time-varying selection bias that can occur in the cohort study. These studies may improve cardiovascular disease prevention for patients with type 2 DM, and this work will lead to proposals for other cardiovascular drug safety studies and launch the career of Dr. Floyd as a leading investigator in this area.